Mechanism analysis of interrupted growth of single-walled carbon nanotube arrays.

Abstract

We investigated the growth mechanism of layered single-walled carbon nanotube (SWNT) mats by a cutting method. Transmission electron microscope observations revealed that new SWNTs grown below first grown SWNTs also have caps at their tips. Raman spectroscopy suggests that the SWNTs in each layer have the same chirality distribution. This growth method might be a way to prove a factor of chirality selection of SWNTs.

Full text

Page 1

MechanismAnalysis of Interrupted
Growth of Single-Walled Carbon
Nanotube Arrays
Takayuki Iwasaki,†John Robertson,‡and Hiroshi Kawarada*,†
Department of Electronic and Photonic Systems, Waseda UniVersity, 3-4-1 Ohkubo,
Shinjuku-ku, Tokyo 169-8555, Japan, and Department of Engineering,
UniVersity of Cambridge, Cambridge CB2 1PZ, United Kingdom
Received November 29, 2007; Revised Manuscript Received January 7, 2008
ABSTRACT
We investigated the growth mechanismof layered single-walled carbon nanotube (SWNT) mats by a cutting method. Transmission electron
microscopeobservationsrevealedthatnewSWNTsgrownbelowfirstgrownSWNTsalsohavecapsattheirtips.Ramanspectroscopysuggests
that theSWNTsineachlayer havethesamechiralitydistribution. Thisgrowthmethodmight beawaytoproveafactor of chiralityselection
of SWNTs.
Carbon nanotubes (CNTs) have been intensively studied
because of their unique structural and electronic properties.
CNTs are candidates for practical applications such as
transistors,1field emitters, LSI interconnects,2,3and capaci-
tors.4,5There is particular interest in nanotubes grown as
vertically aligned mats, both multi-walled carbon nanotubes
(MWNTs)6–10
andsingle-walled
(SWNTs)11–16fabricated by various chemical vapor deposi-
tion (CVD) methods. Some of these applications require
either semiconducting or metallic SWNTs, which depends
on the (n, m) chiral index or wrapping vector of the nanotube.
There are methods to separate the nanotube types after
growth,17,18but they are costly and time-consuming. It would
be preferable to be able to grow a particular type or chirality.
However, this has also been problematic. So far, it has only
been possible to grow a narrow range of chiralities by CVD
because tubes of different chiralities have only a small free
energy difference.
Nanotubes are known to grow by one of two growth
mechanisms, root- or tip-growth, depending on whether the
nanotube grows from the catalyst at its root or tip. Smalley
et. al19,20suggested that a way to produce a nanotube mat of
single chirality is by “cloning”: to form a mat of previously
separated nanotubes and to continue the growth of the
nanotubes by tip growth after “docking” catalyst atoms on
their tips. Here we demonstrate another means to produce
continued growth while retaining chirality distribution, but
in the root growth mechanism, which is the prevalent growth
carbonnanotubes
mechanism of SWNT mats. In addition, we show that this
growth method may prove a factor of chirality selection of
SWNTs.
The layer or marker method was recently used to
determine the growth mode of SWNT and MWNT mats in
CVD.21–24This consists of interrupting the nanotube growth
by lowering the substrate temperature and/or turning off the
plasma, according to some time sequence, and observing
layering of the resulting stack. Growth interruptions produce
clear interfaces. In Figure 1c, it is clear that root growth
occurred because the bottom layer corresponds to the last,
short time growth.
The continued growth experiment raises the question, how
does growth restart? It was found to be possible to separate
the layers at the interface formed by interrupted growth by
cutting by a razor blade. This indicates a weak adhesion
between the layers. This revealed the top surface of the
second, lower layer, which was then characterized. The top
surface of the lower layer was found to possess caps, showing
that it regrew just like a fresh layer, and many selected area
Raman spectra show that the chirality distribution of the two
layers is the same across the interface despite new caps being
formed.
Our vertically aligned SWNTs and double-walled carbon
nanotubes (DWNTs) were grown by a remote plasma CVD
method on Si wafers using a sandwich-like structure catalyst
of 0.5 nm Al2O3/0.3–0.5 nm Fe/5 nm Al2O3. The CVD
system and catalyst preparation are described in detail
elsewhere.21
Layered growth of SWNTs was performed as follows.
First, a substrate with a catalyst was preheated at 600 °C for
* Corresponding author. E-mail: kawarada@waseda.jp. Telephone: +81-
3-5286-3391. Fax: +81-3-5286-3391.
†Department of Electronic and Photonic Systems, Waseda University.
‡Department of Engineering, University of Cambridge.
NANO
LETTERS
2008
Vol. 8, No. 3
886-890
10.1021/nl073119f CCC: $40.75
Published on Web 02/19/2008
 2008 American Chemical Society

Page 2

5 min to convert the Fe film into catalyst nanoparticles. Then
a microwave power of 60 W was applied in a mixture of H2
and CH4gases. The flow rates of H2and CH4were 45 and
5 sccm, respectively, and the total pressure was 20 Torr. The
first layer of SWNTs was grown for 1–5 h, as mats several
hundred micrometers high are needed for cutting. The growth
rate was about 200–300 µm/h. Then the heater and plasma
were switched off and the sample was cooled for 5–30 min,
after which the substrate was heated to 600 °C and the plasma
restarted for growth of the second layer. This procedure was
repeated for more stacks. To make DWNTs, the preheating
temperature was raised to 640 °C to obtain the appropriate
particle size.25The other conditions were the same as that
for SWNT growth.
The resulting nanotubes were characterized by field
emission scanning electron microscopy (FE-SEM), transmis-
sion electron microscopy (TEM), and micro-Raman spec-
troscopy. The Raman spectra were measured by using 514
and 633 nm excitation lasers from cross-sections of each
layer with a 50× objective lens so the area resolution was a
few micrometers.
Parts b and c of Figure 1 show an SEM image of a double-
layered mat. Although layer 2 grew last, it lies below layer
1 because of the root growth mode.21Figure 1c also shows
the interface between layers 1 and 2, with apparently
continuous bundles across the interface.
The layers were separated by applying a lateral force by
a razor blade lightly and slowly to the upper layer, as shown
in Figure 1a. The samples were stuck on glass slides or Si
substrates with double-sided tape. The upper layer would
peel off easily, indicating that adhesion at the interface is
surprisingly weak. This means that the nanotubes may not
be continuous across the interface. Parts d and e of Figure 1
show the sample after cutting layer 1. Only layer 2 was left
on the substrate.
Hand cutting is possible because the upper layers are quite
thick, ∼1 mm.26The cutting process could be optimized;
applying a slicing action from opposite sides produced a
better separation. The top surface of layer 2 has an interesting
structure. Parts f and g of Figure 1 show surfaces of layer 1
before cutting and layer 2 after cutting. It is well-known that
the top surface of CNTs with small diameters such as
SWNTs and DWNTs are curved due to a low density of
CNTs at the initial growth stage.16On the other hand, as
shown in Figure 1g, the top surface of layer 2 shows a
sharper structure, resulting from an almost uniform length
of layer 2 SWNTs.
The nanotube tips of each layer were examined by TEM.
Parts a and b of Figure 2 show tips of a SWNT and a DWNT
in the top layer. Both tips have a cap, and no catalyst particles
were observed. This is typical for root growth.27,28Parts c
and d of Figure 2 show that tips of a SWNT and a DWNT
in layer 2 also have a cap. This is a key observation, as it
shows that regrowth of layer 2 started with new caps, not
by continuing the previous tubes.
Raman measurements were used to measure the diameter
and chirality distribution of the SWNTs in each layer. Figure
3 shows the low frequency radial breathing modes (RBMs)
of layers 1 and 2 for two samples that were grown in different
CVD runs, and the top panel shows a Kataura plot as a
function of nanotube diameter and wavenumber.29The RBM
wavenumber is inversely proportional to the nanotube
diameter. We see that the Raman spectra of layers 1 and 2
for each sample show almost the same RBM peaks (peak
position and relative intensity), while the detailed peak
distribution is different for the two samples.
RBMs of particular nanotube chiralities have a high
intensity when their sub-band gaps are resonant with the
excitation wavelength. This allows the nanotube chiralities
to be indexed from their RBM peaks.30,31We see that
nanotubes in layers 1 and 2 show almost the same peaks in
their RBM peaks in Figure 3. Although the two samples
show different RBM peaks, especially for 633 nm excitation,
the relative intensities in each sample are preserved after
growth of layer 2. This suggests that the two SWNT mats
have the same diameter and chirality distribution in each
layer.
We have three observations: (1) there is weak adhesion
across the interface, (2) new tips occur when growth restarts,
and (3) the diameter and chirality distribution appears to be
continuous across the interface. A possible growth mecha-
nism to account for these factors is shown in Figure 2e. The
catalyst exists as a series of nanoparticles on the support.
First, a carbon cap nucleates on each catalyst nanoparticle
surface, perhaps anchored at its step edges. This cap then
grows into a SWNT in layer 1. When layer 1 growth is
Figure 1. Cutting of layered SWNTs. (a) Schematic of the cutting
procedure using a razor blade. (b,c) Double-layered vertically
aligned SWNTs before cutting. Length of the layer 1 was on the
order of a millimeter for hand cutting. (d,e) The sample after cutting.
Only layer 2 was left on the substrate. The top surface of (f) layer
1 and (g) layer 2.
Nano Lett., Vol. 8, No. 3, 2008 887

Page 3

interrupted, growth restarts by forming a new cap, which
then grows into a tube of layer 2. Because the new cap grows
from the same catalyst nanoparticle, which is solid, it can
have the same diameter as the first cap. The interface between
the new cap and the old tube is not so strong and may be
bound by van der Waals forces. This accounts for the rather
weak adhesion across the interface.
On restarting growth, why do new carbon atoms form a
new cap instead of simply continuing the growth of the
previous nanotube? The answer is that cooling the system
causes a carbon supersaturation within the catalyst nanopar-
ticle. On reheating, this carbon rapidly emerges from the
catalyst and precipitates as a new cap.32–34SWNTs can only
form when there is a moderate, continuous supply of carbon,
not an oversupply.
At an atomic level, a new cap forms if a graphitic cap is
forced to form by rapid precipitation of carbon. This breaks
the normal C-Fe bonds anchoring the cap to the catalyst
surface. If the temperature is not reduced, the C-Fe bonds
are retained and new C atoms can be continuously added to
the root of the growing nanotube without forming a new
cap. The new cap is somewhat analogous to the bamboo
structure that can form in MWNTs.
To show the importance of the cooling on formation of
layer 2, we tried interrupting growth by switching off the
plasma but without cooling. (Note that a plasma is needed
for growth in our system under normal conditions because
the source gas is methane.) Several CVD runs using many
substrates were performed, and no distinct interface was
observed for any samples. During the interval, a catalyst
Figure 2. Investigation of mechanism of layered growth. TEM
images of tip structures of (a) first SWNT, (b) first DWNT, (c)
second SWNT, and (d) second DWNT. (e) Schematic of the growth
mechanism of layered SWNTs. (f) Layered growth of SWNT mats
under different conditions.
Figure 3. RBM peaks in Raman spectra of layers 1 and 2. Raman
spectra were measured using 514 and 633 nm excitation lasers.
Samples 1 and 2 were grown in different CVD runs. The top graph
shows a Kataura plot as a function of nanotube diameter and
wavenumber. Black and red circles indicate semiconducting and
metallic nanotubes, respectively.
888
Nano Lett., Vol. 8, No. 3, 2008

Page 4

remains saturated with carbon but not oversaturated. In this
case, we assume that new carbon atoms of layer 2 precipitate
directly on the edge of the existing SWNTs and continuous
SWNTs are grown. It is interesting to know how small a
temperature change is possible before separation of the layers
occurs. In our case, there might be a slight temperature
change between plasma on and off although the nearly
perfect remote plasma condition, in which there is no
measurable ion current to the substrate, is obtained. However,
this change does not produce the separation. Necessary
cooling temperature will be clarified in future studies by
performing experiments with different temperatures during
the interval between growth of the layers.
The effects of rapid temperature changes on the cluster
shape can be considered. For in situ heating up to 700 °C
and cooling of a catalyst with an as-grown SWNT, Zhu et.
al27found that the catalyst shape and crystallinity were
retained. Thus, the heating for the second growth stage will
not affect the catalyst shape when the same temperature is
used. As a result, the second SWNT has the same diameter
as the first one even though they are not continuous.
Repeating cooling, heating, cap formation, and SWNT
growth should not change the catalyst cluster shape and
crystal phase, and SWNTs with the same diameter can be
synthesized on one cluster.
SWNTs with different diameters grow at higher temper-
atures. When the growth temperature is increased from 600
to ∼700 °C for layer 2 growth, the diameter distribution of
SWNTs shifts to a large diameter,21so that SWNTs grow
with different diameters. It is assumed that catalyst nano-
particles have expanded slightly at the high temperature. This
indicates that diameter and chirality control is possible only
at low temperatures if the shape and steps of the catalyst
nanoparticles are sufficiently well controlled.
We summarize the growth of layered SWNT mats under
different conditions in Figure 2f. No interfaces were observed
if samples were not cooled after the first SWNT growth.
Interfaces were formed only if the sample was cooled after
the first growth. Although new caps were formed at the tips
of the layer 2 SWNTs, SWNTs of both layers grown at the
same temperature had the same chirality distribution. The
two SWNTs are attached by van der Waals attractions so
the layer 1 SWNTs can be easily peeled off from layer 2. If
the growth temperature for the second growth increased,
newly grown SWNTs had larger diameters and different
chiralities, probably due to reshaping of catalyst clusters at
the higher temperature.
It is interesting to consider the growth mechanism in more
detail. How do the layer 2 SWNTs have the same chirality
distribution as those of layer 1? There are two possibilities,
as shown in Figure 4. Figure 4a shows SWNTs with an
identical chirality on a catalyst with a specific crystal phase.
This is the principle of epitaxial control of nanotube
growth.27,28Thus, each particle could create two nanotubes
of the same chirality. The catalyst is solid under our
conditions, as recently demonstrated by in situ studies.35,36
The nanotubes grow by a vapor-solid–solid (VSS) mech-
anism. It is known that the cap structure determines the
chirality of a SWNT. The cap formation strongly depends
on shape and crystallinity of the nanoparticle, probably in
particular, a step-edge structure, which acts as the cap
nucleation site. When two SWNTs grow on the same step-
edge of a nanoparticle, they should have the same chirality.
Another possibility is that SWNTs grow independently of
the crystal phase of nanoparticles, as shown in Figure 4b. In
this case, some SWNTs in layer 2 have different chirality
from layer 1 and chirality is not continuous across the
interface. However, the two layers should have the same
chirality distribution as their Raman RBM peaks show almost
the same relative intensities. In this case, we speculate that
the ratio of SWNTs with specific chiralities would be
determined by growth environment such as temperature,
pressure, and feedstock. This occurs because tubes of
different chiralities have only a small free energy difference.
The diameter of DWNTs is not easily evaluated as SWNTs
because their outer wall diameters (4–5 nm) are below the
range of normal RBM peaks. However, considering results
of the TEM observations, which were similar with SWNTs,
it is expected that DWNTs in layers 1 and 2 have the same
diameter and chirality distribution.
In conclusion, we investigated the mechanism of layered
SWNTs by cutting with a razor blade and found that new
caps were formed when growth restarted. Raman spectra
revealed that the two layers grown at the same temperature
on the set of catalyst particles have almost the same chirality
distribution. Regrowth at a higher temperature caused
different diameters of layer 2 SWNTs. The diameter and
chirality have been preserved in the layer 2 growth, even
after the first process has been terminated when the physical
conditions were kept as same as before. Two possible modes
are proposed. The first one is simply explained by epitaxial
regrowth of SWNTs from an oriented and immobile catalyst
particle. The second one is that physical conditions determine
the distribution of chirality of SWNTs according to the subtle
difference in free energy of chiralities. In this case, irrespec-
Figure 4. Possible combinations of nanotubes with different
chiralities but the same diameter, and catalysts with different crystal
phases in two layers. Catalysts with vertical or horizontal stripes
have different crystal phases but have the same diameter. (a)
Epitaxial growth of SWNTs. (b) SWNT growth independent with
a catalyst crystal phase, but the ratio of specific chiralities is
determined by the growth environment. Pictures of nanotubes were
drawn by software developed by Maruyama.37
Nano Lett., Vol. 8, No. 3, 2008 889

Page 5

tive of orientation of catalyst particle, similar chirality
distribution can be reproduced if the size of particles is fixed.
Acknowledgment. This research was partially supported
by the MIRAI project by NEDO and by “Ambient SoC
Global COE Program of Waseda University” of the Ministry
of Education, Culture, Sports, Science and Technology,
Japan.
References
(1) Javey, A.; Guo, J.; Farmer, D. B.; Wang, Q.; Wang, D.; Gordon, R. G.;
Lundstrom, M.; Dai, H. Nano Lett. 2004, 4, 447.
(2) Nihei, M.; Haribe, M.; Kawabata, A.; Awano, Y. Jpn. J. Appl. Phys.
2004, 43, 1856.
(3) Duesberg, G. S.; Graham, A. P.; Kreupl, F.; Liebau, M.; Seidel, R.;
Unger, E.; Hoenlein, W. Diam. Relat. Mater. 2004, 13, 354.
(4) Li, J.; Cassell, A.; Delzeit, L.; Han, J.; Meyyappan, M. J. Phys. Chem.
B 2002, 106, 9299.
(5) Futaba, D. N.; Hata, K.; Yamada, T.; Hiraoka, T.; Hayamizu, Y.;
Kakudate, Y.; Tanaike, O.; Hatori, H.; Yumura, M.; Iijima, S. Nat.
Mater. 2006, 5, 987.
(6) Fan, S.; Chapline, M. G.; Franklin, N. R.; Tombler, T. W.; Cassell,
A. M.; Dai, H. Science 1999, 283, 512.
(7) Cassell, A. M.; Verma, S.; Delzeit, L.; Meyyappan, M.; Han, J.
Langmuir 2001, 17, 260.
(8) Merkulov, V. I.; Melechko, A. V.; Guillorn, M. A.; Lowndes, D. H.;
Simpson, M. L. Appl. Phys. Lett. 2001, 79, 2970.
(9) Hofmann, S.; Ducati, C.; Robertson, J.; Kleinsorge, B. Appl. Phys.
Lett. 2003, 83, 135.
(10) Christen, H. M.; Puretzky, A. A.; Cui, H.; Belay, K.; Fleming, P. H.;
Geohegan, D. B.; Lowndes, D. H. Nano Lett. 2004, 4, 1939.
(11) Murakami, Y.; Chiashi, S.; Miyauchi, Y.; Minghui, H.; Ogura, M.;
Okubo, T.; Maruyama, S. Chem. Phys. Lett. 2004, 385, 298.
(12) Hata, K.; Futaba, D. N.; Mizuno, K.; Namai, T.; Yumura, M.; Iijima,
S. Science 2004, 19, 1362.
(13) Zhong, G.; Iwasaki, T.; Honda, K.; Furukawa, Y.; Ohdomari, I.;
Kawarada, H. Chem. Vap. Deposition 2005, 11, 127.
(14) Zhang, G.; Mann, D.; Zhang, L.; Javey, A.; Yenilmez, E.; Wang, Q.;
McVittie, J. P.; Nishi, Y.; Gibbons, J.; Dai, H. Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 16141.
(15) Xu, Y. Q.; Flor, E.; Kim, M. J.; Hamadani, B.; Schmidt, H.; Smalley,
R. E.; Hauge, R. H. J. Am. Chem. Soc. 2006, 128, 6560.
(16) Zhang, L.; Tan, Y.; Resasco, D. E. Chem. Phys. Lett. 2006, 422, 198.
(17) Zheng, M.; Jagota, A.; Strano, M. S.; Santos, A. P.; Barone, P.; Chou,
S. G.; Diner, B. A.; Dresselhaus, M. S.; McLean, R. S.; Onoa, G. B.;
Samsonidze, G. G.; Semke, E. D.; Usrey, M.; Walls, D. J. Science
2003, 302, 1519.
(18) Krupke, R.; Heinrich, F.; Lohneysen, H.; Kappes, M. M. Science 2003,
301, 344.
(19) Wang, Y.; Kim, M. J.; Shan, H.; Kittrell, C.; Fan, H.; Ericson, L. M.;
Hwang, W.; Arepalli, S.; Hauge, R. H.; Smalley, R. E. Nano Lett.
2005, 5, 997.
(20) Smalley, R. E.; Li, Y.; Moore, V. C.; Price, K.; Colorado, R.; Schmidt,
H. K.; Hauge, R. H.; Barron, A. R.; Tour, J. M. J. Am. Chem. Soc.
2006, 128, 15824.
(21) Iwasaki, T.; Zhong, G.; Aikawa, T.; Yoshida, T.; Kawarada, H. J.
Phys. Chem. B 2005, 109, 19556.
(22) Li, X.; Cao, A.; Jung, Y. J.; Vajtai, R.; Ajayan, P. M. Nano Lett.
2005, 5, 1997.
(23) Pinault, M.; Pichot, V.; Khodja, H.; Launois, P.; Reynaud, C.; Mayne-
L’Hermite, M. Nano Lett. 2005, 5, 2394.
(24) Zhu, L.; Xiu, Y.; Hess, D. W.; Wong, C. P. Nano Lett. 2005, 5, 2641.
(25) Iwasaki, T.; Maki, T.; Yokoyama, D.; Kumagai, H.; Hashimoto, Y.;
Asari, T.; Kawarada, H. Phys. Stat. Sol. (RRL) 2008, 2, 53.
(26) Zhong, G.; Iwasaki, T.; Robertson, J.; Kawarada, H. J. Phys. Chem.
B 2007, 111, 1907.
(27) Zhu, H.; Suenaga, K.; Hashimoto, A.; Urita, K.; Hata, K; Iijima, S.
Small 2005, 1, 1180.
(28) Reich, S.; Li, L.; Robertson, J. Chem. Phys. Lett. 2006, 421, 469.
(29) Kataura, H.; Kumazawa, Y.; Maniwa, Y.; Umezu, I.; Suzuki, S.;
Ohtsuka, Y.; Achiba, Y. Synth. Met. 1999, 103, 2555.
(30) Telg, H.; Maultzsch, J.; Reich, S.; Hennrich, F.; Thomsen, C. Phys.
ReV. Lett. 2004, 93, 177401.
(31) Maultzsch, J.; Telg, H.; Reich, S.; Thomsen, C. Phys. ReV. B 2005,
72, 205438.
(32) Ding, F.; Bolton, K.; Rosen, A. J. Phys. Chem. B 2004, 108, 17369.
(33) Amara, H.; Bichara, C.; Ducastelle, F. Phys. ReV. B 2006, 73, 113404.
(34) Raty, J. Y.; Gygi, F.; Galli, G. Phys. ReV. Lett. 2005, 95, 096103.
(35) Hofmann, S.; Sharma, R.; Ducati, C.; Du, G.; Mattevi, C.; Cepek, C.;
Cantro, M.; Pisana, S.; Parvez, A.; Cervantes-Sodi, F.; Ferrari, A. C.;
Dunin-Borkowski, R.; Lizzit, S.; Petaccia, L.; Goldini, A.; Robertson,
J. Nano Lett. 2007, 7, 602.
(36) Helveg, S.; Lopez-Cartes, C.; Sehested, J.; Hansen, P. L.; Clausen,
B. S.; Rostrop-Nielsen, J. R.; Abild-Pedersen, F.; Norskov, J. K. Nature
2004, 427, 426.
(37) Maruyama,S.http://www.photon.t.u-tokyo.ac.jp/∼maruyama/
wrapping3/wrapping.html, September 14, 2004.
NL073119F
890
Nano Lett., Vol. 8, No. 3, 2008